Noman-Ahmad25/india-runs-candidate-ranking

🏃 India Runs Candidate Ranking System

AI-powered candidate ranking system built for the India Runs Data & AI Challenge.

Overview

This application ranks candidates using a hybrid approach combining:

• —Feature-based candidate scoring
• —Semantic search with sentence embeddings
• —FAISS vector similarity retrieval
• —Hybrid re-ranking for final candidate selection

Pipeline

Candidates
    │
    ▼
Feature Extraction
    │
    ▼
Embedding Generation
    │
    ▼
FAISS Retrieval
    │
    ▼
Hybrid Ranking
    │
    ▼
Top Candidates

Technologies

• —Python
• —Sentence Transformers
• —FAISS
• —Gradio
• —NumPy
• —Scikit-learn

Demo

Click Generate to run the ranking pipeline and view the highest-ranked candidates.

Repository

GitHub: https://github.com/Noman-Ahmad25/india-runs-candidate-ranking

Notes

This Space uses precomputed artifacts for fast inference:

• —FAISS index
• —Candidate metadata
• —Feature scores
• —Local embedding model

No external APIs are used during ranking.